Workshop opening 16:00 - 18:20

Hayley Francies^{1, 2}

¹ Wellcome Sanger Institute, Cambridge, UK; ² Current affiliation: GSK, Stevenage, UK

Cell lines derived from patient tumours have contributed tremendously to our understanding of cancer biology and therapeutic drug response. However, cancer cell lines grown in 2D in vitro culture have several limitations including failing to represent the heterogeneity of cancer. The derivation of human epithelial organoid models grown in 3D culture could transform the preclinical cancer setting by better reflecting the biology of the tumour of origin, and by providing more predictive models of patient responses to therapies. The Human Cancer Models Initiative (HCMI), an international consortium, intends to derive and characterise the next-generation of cancer cell models. Across the UK, tumour specimens following biopsy or surgery are collected and sent for organoid derivation. Clinical history for each sample is acquired including age, disease stage and prior lines of therapy. Derived organoids are subjected to DNA and RNA sequencing, as well as drug and gene perturbation screens. All models and datasets are made available to the research community via ATCC and web-portals respectively. To date the Sanger Institute has generated and genetically characterised more than 150 cancer organoid models from a variety of cancer types. A new web-portal, the Cell Model Passports has been developed as a mechanism to share clinical as well as genomic and phenotypic datasets for these models. Over the coming years the Sanger Institute hopes to contribute in the region of 600 genomically and phenotypically characterised cell models from a variety of cancer types to the HCMI. We hope these transformative datasets will contribute to the identification of new drug targets and biomarkers of drug response in order to improve patient care.

Tala Tayoun, Françoise Farace
Gustave Roussy, Villejuif, France

Introduction

Metastatic progression is fueled by circulating tumor cell (CTC) dissemination. Nevertheless, most CTCs perish in circulation and only a minor subset survives and harbors tumor-initiating capacities. CTC-derived eXplants (CDX) offer systems for mechanistic investigation of CTC tumorigenicity and biology-driven therapeutic testings. In contrast to patient-derived xenografts, they are established from readily accessible CTC-enriched blood draws. Nevertheless, CDX development remains very challenging mainly due to CTC scarcity in the bloodstream and technical hurdles related to their heterogeneity. In my talk, I will report the establishment and characterization of five CDX models in castration-resistant prostate cancer (CRPC) and non-small cell lung cancer (NSCLC), providing insight into the genetic and mechanistic basis of metastasis and drug resistance in these aggressive malignancies.

Results

Transdifferentiation of CRPC into neuroendocrine (NE) disease CRPC-NE is a recognized resistance mechanism and a major clinical challenge as experimental models are lacking. One CDX and one CDX-derived cell line were developed using CTCs obtained by diagnostic leukapheresis from a CRPC patient at resistance to enzalutamide, which inhibits the AR pathway. Whole-exome sequencing (WES) revealed that the CDX and CDX-derived cell line presented 16% and 56% mutational similarity with the primary tumor (PT) and patient CTCs respectively, while they conserved 83% of PT copy-number alterations, including clonal TMPRSS2-ERG fusion and NKX3.1 loss. Both exhibited an androgen receptor-null phenotype in contrast to the PT and loss of tumor suppressor genes TP53, PTEN and RB1, hallmarks of NE transdifferentiation. While PTEN and RB1 loss were acquired in CTCs, phylogenetic reconstruction suggested that a PT subclone harboring 17p12-tel TP53 loss may have driven the metastatic event leading to the CDX.

DNA damage response (DDR) defects and genomic instability contribute to NSCLC etiology and progression, but their clinical implications remain to be elucidated. 4 CDX models and 3 CDX-derived cell lines were established from NSCLC patient CTCs. WES analysis revealed truncal alterations in key DDR genes across models, which were assessed as therapeutic targets in vitro, in ovo and in vivo. GR-CDXL1 presented homologous recombination deficiency linked to bi-allelic BRCA2 mutation and FANCA deletion, leading to sensitivity to PARP inhibitor (PARPi) olaparib, despite chemoresistance, which challenges the current clinical rationale claiming that chemosensitive NSCLC patients should respond to PARPi. Targeting CIN through centrosome clustering inhibition in GR-CDXL3 impeded tumor growth in ovo and in vivo. In GR-CDXL4, olaparib sensitivity was dictated by SLFN11 overexpression.

Conclusion

Our CRPC and NSCLC CDX and CDX-derived cell lines provide an unprecedented opportunity to understand the biology of CTCs and a robust platform for ex vivo drug testing, for the identification of novel biology-driven therapeutic strategies and clinically useful biomarkers.

Conticelli D.^1,2, Migliore C.^1,2, Picco G.³, Calabrò E.¹, Bellomo S.E.^1,2, Maina I.M.^1,2, Orrù C.^1,2, Ribisi S.^1,2, Corso S.^1,2, Giordano S.^1,2

¹ University of Turin, Dept. of Oncology, Italy
² Candiolo Cancer Institute, FPO-IRCCS, Candiolo, Torino, Italy
³ Wellcome Sanger Institute, Cambridge, UK

Introduction

Around 22% of gastric cancer (GC) cases are characterized by microsatellite instability (MSI), a hypermutability condition caused by the impairment of the DNA mismatch repair (MMR) system. Inactivation of MLH1 or MSH2 genes are the main mechanisms leading to MMR deficiency in both familiar and sporadic MSI GCs. Compared to the other GC molecular subtypes, MSI tumors display a more favorable prognosis, a lower relapse frequency and a better response to immune checkpoint blockade. However, several studies reported the existence, among hypermutated cases, of two populations with different prognosis. Consistently, according to clinical studies testing the anti-PD1 antibody Pembrolizumab alone or in combination with chemotherapy, objective response rates in MSI GCs were close to 60%, suggesting that a significant fraction of tumors are intrinsically resistant. It is thus necessary to get more insights into MSI GC evolution, in order to properly stratify patients, maximize therapy efficacy and identify new cancer vulnerabilities to be exploited.

Material and Methods

To model MSI GC evolution, we established organoid cultures from the normal stomach mucosa of NOD/SCID and BALB/c mice. In line with some reports identifying the MSI status as an early event in gastric carcinogenesis, we generated MMR-deficient (MMRD) organoids using CRISPR/Cas9 technology to knock out MLH1 or MSH2 genes. We maintained both wild type (WT) and MMRD models continuously in culture for several months, evaluated their microsatellite status and molecular features over time, and tested their ability to develop tumor masses upon subcutaneous injection in mice.

Results and Discussion

Our models recapitulate gastric epithelial tissue as they express markers of stem and differentiated gastric cells (i.e. LGR5, MUC5AC, PGC). As expected, after 3 months in culture, MMRD -but not WT models- began to acquire MSI features according to microsatellite marker analysis. Upon subcutaneous injection in NOD/SCID mice, MMRD cells were able to generate tumors in two months. Retrieved cells did not express either Mlh1 or Msh2 and displayed a pronounced MSI phenotype. The evaluation of the Tumor Mutational Burden at several time points is ongoing, as well as the reconstruction of phylogenetic trees recapitulating the evolution of clonal subpopulations. We are also optimizing the model in order to obtain less-immunogenic (non-Cas9 expressing) MMRD BALB/c organoid clones suitable for injection in immunocompetent syngeneic mice to unravel the impact of the immune system and treatment with immune checkpoint inhibitors on tumor evolution.

Conclusion

We used an unbiased approach based on the inactivation of a single MMR gene to generate a powerful model of MSI GC suitable to study tumor evolution, heterogeneity and response to immunotherapy.

Stanislas du Manoir, Hélène Delpech, Béatrice Orsetti, William Jacot, Nelly Pirot, Pierre-Emmanuel Colombo, Claude Sardet, Charles Theillet
IRCM U1194 INSERM, Université de Montpellier, ICM, Montpellier France

Introduction

Most High-Grade Ovarian Carcinomas (HGOCs) are sensitive to carboplatin (CBP)-based chemotherapy but frequently recur within 24 months. Recurrent tumors remain CBP-sensitive and acquire resistance only after several treatment rounds. Recurrences arise from a small number of residual tumor cells hardly amenable to investigation in patients. We developed Patient-Derived Xenografts (PDXs) that allow the study of these different stages of CBP-sensitive recurrence and acquisition of resistance.

Material and Methods

We generated PDX models from CBP-sensitive and intrinsically resistant HGOC. PDXs were CBP- or mock-treated and tumors were sampled, after treatment and at recurrence. We also isolated models with acquired-resistance from CBP-sensitive PDXs. All tumors were characterized at the histological and transcriptome levels.

Results and Discussion

PDX models reproduced treatment response seen in the patients. CBP-sensitive residual tumors contained non-proliferating tumor cells clusters embedded in a fibrotic mesh. In non-treated PDX tumors and treated CBP-resistant tumors fibrotic tissue was not prevalent. Residual tumors had marked differences in gene expression when compared to naïve and recurrent tumors, indicating downregulation of cell cycle and proliferation and upregulation of interferon response and epithelial–mesenchymal transition. This gene expression pattern resembled that described in embryonal diapause and ‘drug-tolerant persister’ states. Residual and acquired-resistance tumors share the overexpression of three genes – CEACAM6, CRYAB, and SOX2.

Conclusion

In HGOC PDX, CBP-sensitive recurrences arise from a small population of quiescent, drug-tolerant, residual cells embedded in a fibrotic mesh. These cells overexpress CEACAM6, CRYAB and SOX2, a signature also associated with acquired resistance and poor patient prognosis, which, thus, might serve as a biomarker to predict recurrence and emergence of resistant disease in CBP-treated HGOC patients.